Non-canonical/planar cell polarity (PCP) Wnt signaling plays important roles in polarizing cells during embryogenesis as well as in human disease (Veeman et al., 2003; Kohn and Moon, 2005; Simons and Mlodzik, 2008; van Amerongen and Nusse, 2009). In vertebrate embryogenesis it mediates morphogenetic movements during gastrulation, neurulation, and head cartilage morphogenesis (Topczewski et al., 2001; Tada et al., 2002; Wallingford et al., 2002; Ciruna et al., 2006). Notably during gastrulation of Xenopus and zebrafish embryos, it is required for convergent-extension movements by promoting medio-lateral cell intercalation. Components of the Wnt/PCP pathway in Xenopus include Wnt11 and Wnt5a, Fz7, Ror2, Disheveled, jun-N-terminal kinase, and small GTPases of the rho/rac/cdc42 family (Djiane et al., 2000; Medina et al., 2000; Tada and Smith, 2000; Habas et al., 2001, 2003; Wallingford et al., 2001; Choi and Han, 2002; Kühl, 2002; Yamanaka et al., 2002; Sheldahl et al., 2003; Schambony and Wedlich, 2007; Semenov et al., 2007). The immediate early responses to Wnt/PCP signaling are changes in the cytoskeleton, cell polarity, protrusive activity, and locomotion of cells. Transcriptional responses to Wnt/PCP signaling have also been observed (Schambony and Wedlich, 2007).
In contrast to Wnt/PCP signaling, immediate early responses to Wnt/β-catenin signaling are typically transcriptional activation of downstream target genes. This permitted the development of transcriptional luciferase reporter assays based on the specific TCF/LEF Wnt/β-catenin response elements (Korinek et al., 1997). These reporters are widely used in the Wnt field since the assay is simple, robust, quantitative, and scalable. The luciferase reporter assays allowed rapid progress in identifying and characterizing the molecular components of the Wnt/β-catenin pathway using gain- and loss-of-function screens. They allow a fast test of whether a gene functions in the Wnt/β-catenin pathway. Using these reporters, the epistasis of Wnt pathway components, i.e., whether they act upstream or downstream relative to each, has been quickly established. They have been used to map the regionalized activity of Wnt/β-catenin signaling in Xenopus and to distinguish between cell-autonomous and non-cell-autonomous signaling by injecting the reporter and candidate effectors in the same or adjacent blastomeres, respectively (Watabe et al., 1995; Wylie et al., 1996; Kiecker and Niehrs, 2001).
Various luciferase reporters for Wnt/PCP signaling have been reported, making use of JNK-dependent AP-1 response elements (Park and Moon, 2002; Le Floch et al., 2005; Lyu and Joo, 2005). However, the response of these reporters to Wnt is modest or their specificity for Wnt/PCP signaling has not been well characterized, and hence they are not widely used. Consequently, most read-outs for PCP signaling rely on semi-quantitative bioassays, such as Drosophila wing hair polarity (Fanto and McNeill, 2004), animal cap elongation (Tada and Smith, 2000), or morphological changes in convergent extension manifesting in changes in marker gene expression as assayed by in situ hybridization (Ewald et al., 2004). Other bio-assays involve live cell analysis of cell protrusive activity (Wallingford et al., 2000), or scanning electron microscope analysis of chochlear hair polarity (Dabdoub and Kelley, 2005). More quantitative biochemical assays are available, such as monitoring JNK phosphorylation or monitoring the activation status of small GTPases (Habas et al., 2001; Yamanaka et al., 2002; Kühl and Moon, 2007). However, these biochemical assays are challenging when working with embryos, because of the low amount of biochemical material available, notably for manipulated embryos.
We therefore searched for a more robust reporter that makes use of the JNK-dependent phosphorylation and activation of transcription factors. JNK is involved in multiple cellular signaling processes and activates the transcription factors jun, fos, and ATF2 (Weston and Davis, 2002). Notably ATF2 was recently shown to be activated by Wnt/PCP signaling during morphogenesis (Schambony and Wedlich, 2007). Here we characterize an ATF2-based reporter and show that it specifically and robustly monitors non-canonical, JNK-dependent Wnt/PCP signaling in Xenopus embryos, which will make it a useful tool in unraveling this signaling cascade in vivo.
RESULTS AND DISCUSSION
Prescreening a variety of JNK-dependent reporters, we identified an ATF2-luciferase reporter (van der Sanden et al., 2004), which faithfully monitored Wnt/PCP signaling in gastrulae. In all subsequent assays, this ATF2-luc reporter was coinjected with a Renilla luciferase reporter plasmid for normalization.
The ATF2-luc reporter was activated by microinjected low doses of Wnt5a mRNA in combination with Fz7 (Fig. 1A), or high doses of Wnt5a or Wnt11 mRNA alone (Fig. 1B), conditions that are known to induce Wnt/PCP signaling in Xenopus (Kim et al., 2008). The reporter was also endogenously activated in Xenopus embryos and this signal was inhibited by previously characterized Morpholinos (Mo) targeting Wnt5a (Fig. 1C and F), Ror2 (Fig. 1D), or Fz7 (Fig. 1E, Winklbauer et al., 2001; Schambony and Wedlich, 2007) or by dominant-negative Wnt11 (Fig. 1F, Tada and Smith, 2000). In contrast, the endogenous and Wnt5a-stimulated ATF2-luc reporter activity were not inhibited by knockdown of LRP6 (Fig. 1C), using a Mo that has been shown to specifically interfere with Wnt/β-catenin signaling (Hassler et al., 2007). In contrast, LRP6 Mo inhibited the Wnt/β-catenin signaling reporter TOPFLASH (Fig. 1G).
We next tested intracellular components of the PCP pathway. Dishevelled (Dvl) plays a critical role in non-canonical Wnt signaling. Hence we tested whether the ATF2-luc reporter was affected by Dvl knockdown using a characterized Mo against Dvl2 (Sheldahl et al., 2003). Injection of Dvl2 Mo reduced endogenous and Wnt5a-stimulated ATF2-luc reporter activity (Fig. 2A). Dvl proteins contain three conserved domains: the N-terminal DIX domain is essential for β-catenin signaling but not for PCP activation, and the central PDZ and C-terminal DEP domains are both required for the PCP function. Dvl deletions lacking the DEP and DIX domain act as dominant negatives for PCP and β-catenin signaling, respectively, and can be used to experimentally distinguish between both pathways (Boutros et al., 1998; Moriguchi et al., 1999; Yamamoto et al., 1999; Heisenberg et al., 2000; Rothbächer et al., 2000; Tada and Smith, 2000; Wallingford et al., 2000; Habas et al., 2001). Consistent with non-canonical Wnt activation, we found that the ATF2-luc reporter is inhibited by ΔDEP- but up-regulated by ΔDIX-Dvl mRNAs (Fig. 2B), probably due to derepression of PCP signalling, which is known to be antagonized by Wnt/β-catenin signaling (Boutros et al., 1998; Moriguchi et al., 1999; Yamamoto et al., 1999; Heisenberg et al., 2000; Rothbächer et al., 2000; Tada and Smith, 2000; Wallingford et al., 2000; Habas et al., 2001). Dvl acts in non-canonical Wnt signaling by activating small GTPases of the rac/rho/cdc42 family (Djiane et al., 2000; Habas et al., 2001). Consistent with this, dominant-negative Rac1 mRNA slightly inhibited, while constitutive active Rac1 mRNA induced, ATF2-luc reporter activity (Fig. 2C). Dominant-negative Rac1 almost abolished Wnt5a-induced ATF2 reporter activity (Fig. 2D). In contrast, injection of dominant-negative CK1γ involved in Wnt/β-catenin signaling (Davidson et al., 2005) or dominant-negative PKC involved in Wnt/Ca2+ signaling (Sheldahl et al., 2003) was without effect (Fig. 2C).
Downstream of the small GTPases, Wnt/PCP signaling is transduced by activation of MKK7 and JNK, which phosphorylate and activate ATF2 (Schambony and Wedlich, 2007). We found that mRNA injection of dominant-negative MKK7 or JNK inhibit endogenous ATF2-luc reporter activity (Fig. 2E).
We conclude that the ATF2-luc reporter monitors endogenous as well as exogenously stimulated Wnt/PCP signaling.
To determine its specificity, we also tested if the ATF2-luc reporter is activated by other signaling pathways relevant during early Xenopus development, including Activin/Nodal, FGF, and BMP signaling (Fig. 3). The ATF2-luc reporter was activated by Activin and constitutively active FGF receptor (caFGFR) but not by constitutive active BMP receptor (CABR). Importantly, both Activin- and caFGFR-induced ATF2-luc reporter stimulations were sensitive to a dominant-negative Dvl for non-canonical pathway (ΔDEP-Dvl, Fig. 3A) or a dominant-negative Wnt11 (Fig. 3B), indicating that Activin and FGF signaling most likely indirectly activate Wnt/PCP signaling, e.g., by inducing the expression of Wnt11 (Tada and Smith, 2000).
Recently, Wnt/PCP signaling has been implicated in the regulation of left-right asymmetry (Aw and Levin, 2009; Antic et al., 2010). This raises the possibility that Wnt/PCP signaling may be asymmetric itself. To test this, we injected the ATF2-reporter in all four blastomeres of 4-cell-stage embryos along with Renilla luciferase for normalization. At gastrula stage (st.12) embryos were bisected sagitally and reporter activity was monitored in left and right halves. In three independent experiments, we detected on average 20% higher ATF-2 reporter activity in right than in left halves (Fig. 4A). This suggests that there is a left-right asymmetry in Wnt/PCP signaling, whose significance remains to be evaluated in future studies.
Finally, we compared ATF2-luc and TOPFLASH reporters and mapped the induction and the regionalization of Wnt/PCP and Wnt/β-catenin signaling. Neither reporter showed significant activity in stage-8–9 embryos (Fig. 4B), consistent with the fact that zygotic transcription only starts at midblastula transition (Newport and Kirschner, 1982). From gastrula stage onwards, both reporters showed progressive activation by endogenous signals, with slightly varying kinetics.
To map their regionalized activity, either TOPFLASH or ATF2-luc reporter plasmid was injected in left and right blastomeres of 8-cell-stage embryos and reporter activity was analyzed at late gastrula stage. Interestingly, ATF2-luc reporter activity was highest in dorso-vegetal blastomeres and decreased ventrally and more so animally, suggestive of a signaling gradient (Fig. 4C, E). The around 8-fold differences between animal-ventral and vegetal-dorsal regions were obtained regardless of the stage or dose of reporter injection (Fig. 5A,B), indicating that the spatial resolution and the sensitivity of the ATF2 reporter in Figure 4 were stage- independent up to 16-cell stage and robust.
TOPFLASH reporter activity was also highest in the dorso-vegetal blastomeres, but showed a greater decrease in animal and ventral blastomeres than ATF2-luc reporter activity (Fig. 4D,E). This distribution of Wnt/β-catenin signaling activity confirms previous observations (Watabe et al., 1995; Wylie et al., 1996) and is consistent with its role in the Nieuwkoop center, which derives from dorso-vegetal blastomeres (White and Heasman, 2008). Likewise, the dorso-vegetal peak of ATF2-luc reporter activity matches the regionalization of convergent extension movements during gastrulation, which are most pronounced on the dorsal side and are driven by Wnt/PCP signaling. The regionalization of ATF2-luc reporter activity may reflect the expression pattern of Wnt11 and Wnt5a, which are expressed in the dorso-equatorial region (Ku and Melton, 1993; Moon et al., 1993; Schambony and Wedlich, 2007). The fact that Wnt/β-catenin and Wnt/PCP signaling are both peaking in dorso-vegetal blastomeres is in accordance with the notion that both pathways synergize in maternal Wnt signaling to promote dorsal axis formation in Xenopus (Cha et al., 2008).
In conclusion, we describe a luciferase reporter, which faithfully monitors Wnt/PCP signaling in Xenopus. While we have not tested the utility of the reporter in embryos of other species, we note that the reporter was only slightly activated by Wnt5a in various mammalian cell lines tested (not shown). The reason for this difference between Xenopus and the mammalian cell lines is unknown, but clearly indicates context dependence of the reporter, possibly due to lack of required transcriptional cofactors. Nevertheless, this ATF2-luc reporter will aid characterizing novel signaling components involved in non-canonical Wnt signaling in Xenopus and possibly other vertebrate embryos. It may be interesting to study what the nature of the context-dependence for this reporter is. It may also be interesting to generate transgenic animals in which ATF2 drives GFP expression, to monitor non-canonical Wnt signaling in live embryos.
mRNA including full-length of Xenopus Fz7, Wnt5a, and Wnt11 were used for injections. In Xenopus ck1 gamma mutant constructs (dnCK1) lysine 73 were replaced by arginine and used as dominant negative (Davidson et al., 2005). Xenopus Wnt11 mutant construct (dnWnt11), Xenopus dishevelled mutant constructs (ΔDIX and ΔDEP), Xenopus JNK and MKK7 mutant constructs (dnJNK and dnMKK7), Xenopus PKC and Rac mutant constructs (dnRac1, caRac1, and dnPKC), constitutive active Xenopus FGF receptor (caFGFR), and Xenopus constitutive active BMP receptor (CABR) were gifts from M. Tada (dnWnt11, ΔDIX, and ΔDEP), Y. Sasai (dnJNK and dnMKK7), R. Mayor (dnRac1, caRac1 and dnPKC), K.M. Neilson (caFGFR), and K.W. Cho (CABR).
In vitro fertilization, embryo culture, preparation of mRNA, and microinjection were carried out as described (Gawantka et al., 1995). Equal amounts of total mRNA or Mo were injected by adjustment with the preprolactin (PPL) mRNA or the standard control Mo (Gene Tools, Philomath, OR), where necessary. For left-right reporter assays, devitellinated gastrula stage (st.12) embryos were placed on agarose-lined dishes with the ventral side facing up. Embryos were bisected longitudinally along the AP axis with a hair knife, with the closing blastopore serving as posterior indicator.
Luciferase Reporter Assays
For luciferase reporter assays in Xenopus embryos, embryos were injected with ATF2-luc (van der Sanden et al., 2004) or TOPFLASH and Renilla-TK plasmid DNA plus Morpholinos and/or synthetic mRNA indicated. Three pools of 7 embryos each were lysed with passive lysis buffer (Promega, Madison, WI) and assayed for luciferase activity using the Dual luciferase system (Promega). All luciferase reporter assays represent the mean ± standard error of the mean (SEM) from 3 independent measurements of pools (7 embryos per pool; total n=21 per experiment shown). The reproducibility was confirmed by at least three independent experiments in different batches of Xenopus laevis and/or Xenopus tropicalis embryos.
We thank E. Karaulanov for technical advice and conceptual input and H. Steinbeisser and P. Angel for providing reagents.